Oral Presentations—Methods and Materials of Cleaning, Conservation, Repair and Maintenance; Logistics and Planning

8/12/2019 Oral PresentationsMethods and Materials of Cleaning, Conservation, Repair and Maintenance; Logistics and Plan

1/41

12

th

International Congress on the Deterioration andConservation of Stone

Wednesday 24 October 2012

Oral PresentationsMethods and Materials of Cleaning,

Conservation, Repair and Maintenance; Logistics and

Planning

Session IX: 10:30 12:15


2/41

CHARACTERIZATION OF HYDRAULIC MORTARS CONTAINING NANO-

TITANIA FOR RESTORATION APPLICATIONS

N. Maravelaki1*

, E. Lionakis1, C. Kapridaki

1, Z. Agioutantis

2, A. Verganelaki

1, V.

Perdikatsis2

1Department of Science

2Department of Mineral Resources Engineering1,2Technical University of Crete, Polytechnioupolis, Akrotiri, 73100 Chania, Crete,

Greece

Abstract

In this work nano-titania of anatase form has been added in mortars containing (a):binders of either lime and metakaolin or natural hydraulic lime and, (b): fine aggregates

of carbonate or silicate nature. The aim was to study the effect of nano-titania in thehydrolysis and carbonation of the above binders widely used in the design of restorationmortars, as well as the mechanical properties of the derived mortars. The nano-titania

proportion was 4.5-6 per cent w/w of binders. The physicochemical and mechanical

properties of the nano-titania mortars were studied and compared to the respective ones,without the nano-titania addition, used as reference. DTA-TG, FTIR and XRD analyses

indicated the evolution of carbonation, hydration and hydraulic compound formation

during a one-year curing. The mechanical characterization indicated that the mortarswith the nano-titania addition showed improved mechanical properties over time when

compared to the specimens without nano-titania. The results evidenced carbonation and

hydration enhancement of the mortar mixtures with nano-titania. The hydrophylicity ofnano-titania improves the humidity retained in mortars, thus facilitating the carbonation

and hydration process. This property can be exploited in the fabrication of mortars for

adhering fragments of porous limestones from monuments, where the presence of

humidity controls the mortar setting and adhesion efficiency. A specifically designedmechanical experiment based on the direct tensile strength proved the suitability of these

mortars with nano-titania as adhesive materials for restoration applications.

Keywords: adhesive mortars, nano-titania, metakaolin-lime, hydraulic lime, hydration,mechanical properties

1. IntroductionThe adhesion of mortars to fragments of archaeological stone or other buildingmaterials is an important intervention, which results in a substantial structural integrity

between the adhered materials, leading to the slowing or preventing from further decay.Treatment options include the application of adhesives and grouts, as well mechanical

pinning repairs. Commonly used adhesives such as epoxy, acrylic and polyester resins

demonstrated excessive strength, high irreversibility and, if improperly applied, theirremoval may be more damaging to the historic fabric (Amstock 2000).

*Corresponding author. Tel.: +30 (28210) 37661; fax: +30 (28210) 37841.

E-mail address:[email protected](Noni Maravelaki)
mailto:[email protected]:[email protected]:[email protected]:[email protected]


3/41

In this research work, two different kinds of stone from Piraeus, namely Aktites and

Mounichea stones corresponding to a hard dolomitic limestone and a marly limestone of

the area, respectively, were selected due to their common employment as main

construction materials of the Athenian Acropolis building during the Archaic period.

The preferred treatment strategy for the reassembling and adhesion of these fragmentswas addressed through designing bonding mortars compatible to these stones. Repair

criteria, were as follows: (a) physico-chemical and mechanical compatibility between

repair materials and stone, (b) adequate strength to resist tensile and shear forces, (c)retreatability, (d) longevity, (e) affordability and (f) ease of installation.

The design of adhesive mortars with binders of either hydrate lime-metakaolin or

natural hydraulic lime has been adopted with the aim of formulating a complex systemcharacterized by the highest compatibility. Nowadays, both of hydrate lime-metakaolin

and hydraulic lime mortars are widely used in the field of the restoration andconservation of architectural monuments, due to its capability to enhance the chemical,

physical, structural and mechanical compatibility with historical building materials(stones, bricks and mortars) (Rosario, Velosa, Magalhaes, 2009). This compatibility is avery critical prerequisite for the optimum performance of conservation mortars

considering the damages caused to historic monuments during the past decades, due to

the extensive use of cement mixtures.Into this framework, in special designed mortars consisting of binders of either lime

and metakaolin or natural hydraulic lime and fine aggregates of carbonate nature, nano-

titania of anatase (90 per cent) and rutile (10 per cent) has been added in 4.5-6 per cent

w/w of binder. The aim was to study the effect of nano-titania in the hydration andcarbonation of the above binders and to compare the physico-chemical properties of the

nano-titania mortars with those mortars without nano-titania (used as reference).

Thermal analysis (DTA-TG), infrared spectroscopy (FTIR) and X-ray diffraction (XRD)

analyses were performed to investigate the evolution of carbonation, hydration andhydraulic compound formation during a six-month curing period. Furthermore, the

stonemortar interfaces, the adhesion resistance to external mechanical stress as related

to the physico-chemical characteristics of the stone-mortar system and the role of the

nano-titania as additive were reported and discussed in this paper.

2. Experimental procedure2.1 Design of adhesive mortars: binders, fillers and aggregates

Binders of either lime (L: by CaO Hellas) with metakaolin (M: Metastar 501 byImerys), or natural hydraulic lime (NHL: NHL3.5z by Lafarge) as well as nano-titanium

dioxide (T: nano-structured nano-titania by NanoPhos), used as filler due to its

photocatalytic activity, are employed for the design of the adhesive mortars. The alreadyestablished photocatalytic activity of nano-titania in anatase form (Hyeon-Cheol,

Young-Jun, Myung-Joo et al. 2010) will significantly enhance the hydration and

carbonation process, thus affecting the adhesion performance. Moreover, self-cleaningproperties of the adhesive mortars can be also attained due to the photocatalytic action

of the nano-titania. XRD, FTIR and DTA-TG techniques were used to characterize the

raw products.

In Table 1, the mortar mixes are presented, where the ratio of water to binder (W/B)ranges from 0.8 to 0.6. The required quantity of lime that will react with metakaolin was


4/41

fixed in a weight ratio equal to 1.5, ensuring the pozzolanic reaction. Any unreacted

quantity of lime, after its carbonation, provides elasticity to the final mortar and enables

the mortar to acquire a pore size distribution similar or compatible to porous stone, thus

allowing a homogeneous distribution of water and water vapor in the complex system.

Furthermore, the enhanced derived elasticity can function as a tool for the arrangementand absorption of external stresses, which otherwise could lead to the mechanical failure

of the mortar.

Table 1: Mortar mixes (composition in mass %)Samples Sand Binders Filler B/A W/B

Code Cc NHL M L Nano-titania

NHLT1 48 49 3 1 0.7

NHLT2 33 64 3 2 0.6

ML1 50 20 30 0 1 0.8

MLT1 47 20 30 3 1 0.8

Preliminary tests on the adhesive capability of the designed mortars with fragments

of porous limestones pointed out the inefficient performance of NHL mortars. Thus itwas decided to exclude this formulation from further study. TiO2 nano-powder

dispersion in a small amount of water was achieved through ultrasonic treatment for 15min; afterwards the obtained TiO2colloidal solution was subjected to UV radiation (365

nm) for 30 min to activate the nano-titania. Then the dispersed TiO2solution was mixed

with the other raw material and stirred with a handheld mixer for 5 min. Due to the factthat the fine aggregates can contribute to the avoidance of shrinkage and cracking during

the setting process, the addition of sand with fine grains was deemed essential.Consequently, equal proportions of sandpassing through the 125 and 63 m sieves wereadded in the mix, which were previously washed by water to free the harmful soluble

salts.

2.2 Assessment of the adhesive mortars2.2.1Physico-chemical properties of mortars

When binders of powdered pozzolans, such as metakaolin are mixed with lime, or

natural hydraulic lime is mixed with water, they produce a new binder that exhibits a

hydraulic character due to reactions among the amorphous phase of pozzolans and lime(Aggelakopoulou, Bakolas and Moropoulou 2011), as well as hydration of NHL(Maravelaki-Kalaitzaki Karatasios Bakolas et al. 2005). The pozzolanic and hydration

reactions, which take place in room temperature and in conditions of high relative

humidity, lead to the formation of a hydrous gel of calcium silicate hydrate (CSH) and

calcium aluminate hydrated phases (CAH), which modify the microstructure of the pasteand increase both the hydraulic properties and the strength of the mortar (Tziotziou

Karakosta Karatasios et al. 2011). Therefore, the study of the hydration is essential in

order to evaluate the performance of the mortar, in terms of physical and mechanicalproperties, which are also interrelated to the longevity of the mortar (Papayianni and

Stefanidou 2006).


5/41

The above mixtures (Table 1) were molded in prismatic and cubic moulds, with

dimensions of 4416 cm and 555 cm, respectively and then placed in a curing

chamber for setting, at RH = 65 3 % and T = 20 2 oC, according to the procedure

described in the EN 196-1 standard. Pastes of these mixtures with and without nano-

titania, with dimensions of 5 mm in diameter and 30 mm in height, were also preparedand sealed into ceramic tubes using Parafilm membrane to avoid moisture loss and

drying and were then maintained at the same curing conditions with the studied mortars.

The setting process of the paste was interrupted at preset time periods, of 1, 3, 5, 7, 11,21, 28 and 90 days according to a hydration stop procedure, which involved the

immersion of the sample in two stop-bath solutions (acetone and diethyl-ether) for 60

min each, and then drying at 70oC for 30 h.

The development of the hydration and carbonation of powder samples were carried

out by XRD, mercury intrusion porosimetry (MIP), FTIR, DTA/TG, EDXRF andscanning electron microscopy (SEM). By identifying CSH and CAH at different ages,

not only qualitatively, but also semi-quantitatively, the hydration and carbonationprocess can be monitored. The mineralogical analysis was investigated by XRD with aSiemens D-500 diffractometer (40 kV/35 mA) and the spectra were collected between 5

and 60o 2 scale, with a step of 0.03/5s. SEM analysis was carried out in fractured

surfaces, using a FEI Quanta Inspect scanning electron microscope. DTA/TG wasoperated with a Setaram thermal analyser; in static air atmosphere up to 1000 oC at a

rate of 10 oC/min. The FTIR analysis of KBr pellets was operated in Perkin-Elmer

spectrophotometer in the spectral range of 400-4000 cm-1. MIP measurements wererecorded using a Quantachrome Autoscan 60 porosimeter, in the range of 2-4000 nm.

Physical properties of the stone and mortars were further studied by water absorption by

saturation according to EN 13755:2002, as well as capillary water absorptionmeasurements for mortars according to EN 1015-18:1995 and for stones according to

EN 1925:1999.

2.2.2Mechanical estimation of the designed adhesive mortarsThe Aktites and Mounichea stone samples were cut and shaped into specimens with

dimensions of: (a) 4x4x4 cm used to measure the compressive strength of the stone and(b) 4x4x8 cm used to bond them with the designed mortars. The mechanical properties

of the designed mortars were characterized by measuring the uniaxial compressive

strength (Fc) and the flexural strength (Ff-3pb) according to EN 1015-11:1999.The incising of stone faces by special mechanical tools to provide a rough surface

and the wetting of these surfaces, were prerequisite before the application of the mortars.The mortar was partially applied to the stone surface and then the stone specimens were

filled with mortar by placing them levelly with the aid of special joint clamps. The jointwas kept moist with damp cotton gauze and a polyethylene sheet. The specimens were

then placed in a storage chamber for 28 days, according to the conditions described in

EN 1015-11:1999.

Home designed equipment for both the four point flexural strength (Ff-4pb) (Fig. 1a,b) and the direct tensile strength test (Ft) (Fig 1c), were used for measuring the adhesive

performance of the bonded stone-mortar specimens (4x4x17 cm) (Whittaker, Singh and

Sun 1992).


6/41

Figure 1. a) Four point bending apparatus with variable support spans b) geometry andcalculations for four-point bending test c) direct tensile test apparatus for stone

mortar specimens.

3. Results and discussion3.1 Stones and raw materials characterizationTable 2: Mineralogical composition and properties of the stones and raw materials.

Stone Cc Do Cl Qz Kl Il Alb Fc(MPa)

WCC(g cm-1s-1/2)

WSC%

D1 6.2 83.0 0.6 3.7 1.8 4.2 0.5 42.3 0.0026(0.001)

3.7(0.1)

D2 1.5 78.7 2.8 6.1 2.5 7.8 0.5 28.7 0.0141(0.004)

13.1(2.8)

D3 12.0 80.0 0.9 2.7 2.4 1.9 0.1 107.0 0.0011(0.0002)

2.3(0.2)

D4 4.2 80.0 1.3 3.3 3.3 7.0 0.9 6.6 0.0007

(0.0004)

2.9

(0.9)K 11.1 70.9 1.1 5.4 0.9 8.1 2.2 22.7 0.0211

(0.005)

8.3

(2.0)Cc: Calcite; Do: Dolomite; Cl: Chlorite; Qz: Quartz; Kl: Kaolinite; Il: Illite; CH: Calcium

Hydroxide; C2S-beta: Larnite; C3S: Alite; At: Anatase; Rt: Rutile; Fc: Compressive Strength;

WCC: Water Capillary Coefficient; WSC: Water Saturation Coefficient

Mineralogical composition and properties, both of raw materials and stones,

determined by XRD and thermal analyses are presented in Table 2. According to these

results, the Piraeus stone can be classified in: (a) micritic dolomitic limestone hard andcompact, with a low to medium porosity, small grain size and high values of mechanical

strength (D1, D3); (b) marly limestone/dolomitic limestone with an oolitic texture,

brownish or yellowish to light grey-color with a low to medium porosity and low valuesof mechanical strength (D2, D4, K).

Furthermore, Table 2 reports the values of the compressive strength, as well as

mean values of water capillary and water saturation coefficient calculated in three stonespecimens. The water capillary and water saturation tests indicate that the pore system


7/41

of the studied stones differed significantly and therefore the designed mortars should be

adapted accordingly. Stones with high compressive strength, such as D1 and D3 absorb

a low water amount correspondingly to their low quantity of aluminosilicates. However,

in the D2, D4 and K stones no clear relationship exists between compressive strength

and hygric properties. Even though these stones contained similar amounts ofaluminosilicates, they differed both in the absorbed quantity of water and the values of

the compressive strength. This relies on the structural inhomogeneity of the samples and

especially, as in the case of D4 stone, on the absence of inter-connected pores, whichaffect the stone hygric behavior.

3.2Physico-chemical evaluation of mortarsThe total bound water (Htb) identified in the studied samples in the temperature

range from 100 to 460oC corresponds to the dehydration of calcium silicate hydrates

(CSH), calcium aluminate hydrates (CAH) and the residual bound water

(Aggelakopoulou, Bakolas and Moropoulou 2011). The dehydration of Ca(OH)2 (CH)occurred in the temperature range ~480 500

oC, while the decomposition of CaCO3

took place at a temperature higher than 600 oC.

Figure 2. DTA curves for NHL mortars without nano-titania (dashed line) and with nano-titania (solid line) at 7 days of curing along with the evolution of carbonation

illustrated in the inset plot.

In particular, Figure 2 depicts the DTA curves for the NHL mortars with and

without nano-titania at 7 days of curing. The inset plot illustrates the evolution of theratio unreacted-CH (CHun) to formed-Cc(Ccf) for a curing period of 1, 3, 5 and 7 days.

As the setting process proceeds mass losses attributed to the release of CO2from CaCO3

increased, while the mass loss of CH dehydration decreased, due to the transformationof CH into hydraulic components and calcite. The lime consumption and the formation

of hydraulic phases are more pronounced for the nano-titania mortars at different curing

times. The same observation was also obtained from the thermal analysis of metakaolin-

lime mortars with and without nano-titania.

SEM micrographs of the mortars ML1 without- (Figure 3a) and with nano-titaniaMLT1 (Figure 3b) after one year of curing corroborate the results of DTA analysis.


8/41

SEM micrograph of MLT1 (Figure 3b) shows that a dense network of hydraulic

amorphous components was formed providing more elasticity to the mortar matrix.

Moreover, characteristic hexagonal crystals of portlandite were located in both SEM

micrographs; nevertheless in the case of MLT1 mortar the portlandite crystals are

obviously fewer. The FTIR spectra of the above studied mortars are in full accordancewith the SEM micrographs, showing after curing of one year traces of portlandite and

enhanced hydraulic compound formation (Maravelaki-Kalaitzaki, P. Lionakis, E.

Agioutantis, Z. et al.2012).

Figure 3. SEM micrographs of (a) ML mortar and (b) MLT mortar showing portlanditecrystals (P) and the hydraulic amorphous components.

These variations can be explained by the TiO2photocatalytic action for the mixtures

with nano-titania which leads both to calcite formation and enhanced hydration of CSHand CAH products, similar to what was observed on the early age hydration of Portland

cement by adding increased dosage of fine titanium oxide (Jayapalan, Lee and Kurtis

2009).

3.3Mechanical EvaluationTable 3 reports the physical and mechanical properties of the designed mortars. The

physical properties of the designed mortars differed insignificantly indicating that thenano-titania addition neither modified the microstructure nor affected the hygric

behaviour of the materials. The lowest Fc values were recorded for the NHL samples,

while the Fc values decreased with curing time in the ML1 samples. Even though, the Fcvalues recorded at four weeks curing for the MLT1 samples are lower than the

corresponding values for samples without nano-titania (ML1), nevertheless, the Fc

values of the MLT1 samples reached higher values than the ML1 samples after threemonths and one year curing, thus indicating the beneficial effect of the nano-titania in

the compressive strength. The decrease of Fc values over time in the ML1 compositions

has been already reported by other authors (Aggelakopoulou, Bakolas and Moropoulou2011, Velosa Rocha and Veiga 2009) and was most probably attributed to the

microcracking formation due to shrinkage during the curing. Figure 4 depicts typical

stress-strain curves of hydrated lime-metakaolin with (MLT1) and without nano-titania(ML1), in two time intervals of 4 weeks and 3 months.


9/41

Figure 4. Stress-strain diagram of sampleswithout nano-titania at 4 weeks (ML1 [4w]) and

3 months (ML1 [3m]) curing time and with nano-titania at 4 weeks (MLT1 [4w])

and 3 months (MLT1 [3m]) curing time.

Table 3: Physical and chemical properties of the designed mortars.Code WCC*

g cm-1

s-1/2

WSC*

%

P

%

Pr

m

SSA

m2/g

Cur

-ing

Fc^

MPa

Ff-

3pb*

MPa

E

GPa

NHLT1 0.0149

(0.007)

27.80

(1.09)

32.96 0.295 12.0 4w

3m1y

4.15(0.56)

5.47

(0.28)5.51 (0.95)

-

1.69-

0.17

NHLT2 0.0080

(0.0004)

25.58

(0.27)

37.79 0.092 12.84 4w

3m

1y

5.41(0.29)

7.22 (1.61)

-

-

-

0.37

ML1 0.0040

(0.0001)

33.92

(0.26)

31.46 0.031 14.02 4w

3m

1y

14.85(1.53)

11.57 (1.39

10.62(2.39)

1.15

-

-

0.42

0.76

0.56

MLT1 0.0074

(0.003)

29.52

(1.07)

32.54 0.031 16.01 4w

3m

1y

9.08* (0.89)

14.19(0.70)

15.40(0.70)

1.21

-

-

0.59

1.10

0.91

(*) mean value of three samples; (^) mean value of 6 samples; Fc: compressive strength; Ff-3pb: Flexuralstrength; E: Elasticity modulus, [4w]: 4 weeks; [3m]: 3 months; [1y]: 1 year; -: n. a.

By combining the results of physical and mechanical properties the mortars NHLT2,ML1 and MLT1 were selected as adhesive means for the stones under consideration

(Table 4). The mortar NHLT1 showed similar values to others except for its high water

coefficient capillary, which can be considered of secondary importance for the adhesiveability. However, NHLT1 exhibited difficulty in joining the stone specimens and

therefore was not included in the finally selected mortars.

It seems that nano-titania with its hydrophylicity created an environment, which notonly enhanced the hydraulic component formation, but also controlled the shrinkage,

thus avoiding microcracking (Karatasios Katsiotis Likodimos, et al. 2010). Further


10/41

support to this statement is derived from the dense network of hydraulic components

observed in the SEM micrographs (Fig. 3) for the MLT1 samples.

Table 4: Mechanical properties of stone-mortar specimens cured for four weeks.Mortar

Code

Number of Ff-4pb

adhered stone

specimens

Ff-4pb

(MPa)

Number of Ft

adhered stone

specimens

Ft

(MPa)

NHLT2 4 (D1, D2) 2.39 (0.7) 2 (D1, D2) 0.51 (0.26)

MLT1 3 (D3, K) 1.34 (0.47) 2 (K) 0.15 (0.08)

ML1 1 (D4) 1.07 1 (D1) 0.09

Table 4 reports the results of the 4-point bending test and the direct tensile test for

the adhered stone-mortar specimens. In all tests, failure was observed at the interface

between stone and mortar. The results revealed that the higher bonding strength (higher

flexural and tensile) was measured when applying the NHLT2 mortar as compared tothe MLT1 and ML1 mortar.

4. ConclusionsThis research work addressed an important problem in the restoration sector

concerning the reassembling of stone fragments from ancient monuments using non-cement mortars. The proposed adhesive mortars contain hydraulic lime or metakaolin

and lime as binders, carbonate sand with grains smaller than 250 m in a B/A ratio from

1 to 2, as well as nano-titania as additive in a binder replacement of 4.5-6%.The mechanical characterization indicated that the mortars with nano-titania showed

increased compressive and flexural strength and modulus of elasticity when compared to

the specimens without nano-titania. The results also indicate enhanced carbonation and

hydration of mortar mixtures with nano-titania. The hydrophylicity of nano-titania

improves the humidity retention into mortars, facilitating thus the carbonation andhydration processes. This property can be exploited in the fabrication of mortars tailored

to adhering porous limestones, where humidity controls the mortar setting and adhesion

efficiency. Home-designed equipment applied to measure the bonding strength of stone-

mortar systems revealed that the nano-titania addition in both metakaolin-lime andhydraulic lime mortars improves the adhesive property of the mortar when applied to

porous stones.

ACKNOWLEGMENTS

This work was carried out in collaboration with the The Akropolis Restoration Service

(YSMA) of the Hellenic Ministry of Culture and Tourism. The authors would like to

thank Emeritus Prof. Vasileia Kasselouri-Rigopoulou, (Committee for the Restoration ofthe Acropolis Monuments-ESMA) for collaboration and scientific support, the

Committee for the Restoration of the Acropolis Monuments and Dr E. Aggelakopoulou

Head of the Technical Office for the Acropolis Monuments' Surface Conservation. This

research has been co-financed by the European Union (European Social Fund - ESF)and Greek national funds through the Operational Program "Education and Lifelong


11/41

Learning" of the National Strategic Reference Framework (NSRF) - Research Funding

Program: Heracleitus II, Investing in knowledge society through the European Social

Fund.

5. ReferencesAggelakopoulou E, Bakolas A, Moropoulou A. 2011. Properties of limemetakolin

mortars for the restoration of historic masonries.Applied Clay Science, 53(1):1519.

Amstock JS. 2000. Handbook of adhesives and sealants in construction. New York:McGraw-Hill.

Hyeon-Cheol J, Young-Jun L, Myung-Joo K, Ho-Beom K, Jun-Hyun H. 2010.

Characterization on titanium surfaces and its effect on photocatalytic bactericidalactivity.Applied Surface Science, 257(3):741746.

Jayapalan AR, Lee BY, Kurtis KE. 2009. Effect of Nano-sized Titanium Dioxide on

Early Age Hydration of Portland Cement. In Nanotechnology in Construction,

Proceedings of the NICOM3.Karatasios I, Katsiotis MS, Likodimos V, Kontos AI, Papavassiliou G, Falaras P,

Kilikoglou, V. 2010. Photo-induced carbonation of lime-TiO2 mortars. AppliedCatalysis B: Environmental, 95(1-2):78-86.

Maravelaki-Kalaitzaki, P. Lionakis, E. Agioutantis, Z. Kparidaki, C. Verganelaki, A.

Mayrigiannakis, S. Stavroulaki, M. Perdikatsis, V. Kallithrakas-Kontos, N. 2012.Physico-chemical and mechanical characterization of hydraulic mortars containing

nano-titania for stone applications, 4th International Symposium on

Nanotechnology in Construction, Agios-Nikolaos.Papayianni I, Stefanidou M. 2006. Strengthporosity relationships in limepozzolan

mortars. Construction and Building Materials, 20(9):700-705.Rosario Veiga M, Velosa A, Magalhaes A. 2009. Experimental applications of mortars

with pozzolanic additions: Characterization and performance evaluation.Construction and Building Materials, 23(1):318-327.

Tziotziou M, Karakosta E, Karatasios I, Diamantopoulos G, Sapalidis A, Fardis M,

Maravelaki-Kalaitzaki P, Papavassiliou G, Kilikoglou V. 2011. Application of 1H

NMR in hy-dration and porosity studies of lime pozzolan mixtures. Microporousand Mesoporous Materials, 139(1-3):16-24.

Velosa AL, Rocha F, Veiga R. 2009. Influence of chemical and mineralogical

composition of metakaolin on mortars characteristics. Acta Geodynamica etGeomaterialia, 153(6):121-126.

Whittaker B., Singh N., Sun G. 1992.Rock Fracture Mechanics: Principles, Design andApplications, Elsevier.


12/41

Compatibility of Roman cement mortars with gypsum stones and anhydritemortars: The example of Valre Castle (Sion, Switzerland)

C. Gosselin1*, F. Girardet2,S.B. Feldman3

1 Laboratory of Construction Materials, Ecole Polytechnique Fdrale de Lausanne, CH-1015 Lausanne,Switzerland, [email protected] Sarl, CH-1807 Blonay, Switzerland, [email protected] Minerals International LLC, Cockeysville, USA, [email protected]

1.AbstractDurable and reversible restoration of stones and historical mortars is a major concern to thoseinterested in conservation of historical structures and contemporary practices of restorationare continuously revisited with the help of the technical and scientific researches. However,the historical feedback on repairing techniques recently showed that Roman cements (RC),

developed and widely used through the XIXth Century, were particularly well-adapted torepair historical masonries. The current article presents a case study of RC mortars applied ongypsum stones and historical anhydrite mortars, both soluble and known to be sensitive to thechemical compatibility with hydraulic binders. Mineralogical analysis of samples from the

basilica Notre Dame de Valre (Sion, Switzerland) shows that late XIXth C. RC joints andrenders have perfectly lasted in contact with the structural gypsum stones and anhydritemortars from the XIIIth C. Results from XRD and SEM work suggest that the present RC was

produced at a temperature high enough to form significant amounts of -C2S and C2AS,remaining unreacted after very long term hydration. The extent of C2S hydration is notablyreduced due the precipitation of silica gel, a carbonation product, at the boundary of thecement grains. The high capillary porosity developed during hydration is homogeneouslydistributed, enhancing the transport properties. These conclusions were supported bycomplementary observations. First, elemental mapping through the strong RC /anhydritemortars interface does not indicate any accumulation of sulfate salts at the boundary.Additionally, in contrast to the RC mortars, the rapid expansion and degradation of greyPortland cement mortars was observed, confirming the limitations of the latter applied ongypsum stones.

Keywords: Roman cement, gypsum stone, anhydrite mortars, microstructure, compatibility

* Now at Geotest, Le Mont-sur-Lausanne, Switzerland, [email protected]

2.IntroductionThe basilica Notre Dame de Valre (also called Castle of Valre) is part of a fortified village

built in the XIIth C., on the top of a hill overlooking the city of Sion. In 1877, the cantonalGreat Council called upon the Government to report on the ownership of the feudal castles ofthe canton and the measures for their conservation. A major restoration campaign of theValre Castle was conducted between 1892 and 1902 under the supervision of the SwissSociety for the Conservation of Historic Buildings. The work was led by the OfficeKalbermatten under the direction of the architect Theophile van Muyden [1].


13/41

The masonry is composed of different types of local stones (calcareous, tuff and gypsumstones). Two varieties of local gypsum stones were used (white gypsum stones, i.e. alabaster,and yellowish gypsum stones containing sand) for both ashlar masonry and sculptedornamentations in areas unexposed to rain. The protection of the gypsum stones from the rainled to their relatively good state of conservation despite their high solubility (2 g/l). However,

the dissolution of gypsum stones in areas exposed to rain, and the subsequent release ofsulphate ions can lead to major conservation concerns related to their compatibility with thedifferent hydraulic binders used for the joints, the structural and decorative elements [2]. Thisarticle discusses the specific case of restoration joints made of XIXth C. Roman cement basedmortars.

Several types of mortars are present on the faades, according to the period of constructionand restoration of the building. Six types of mortars were identified as shown in Table 1. The

present study focuses on the compatibility of the XIXth C. Roman cement mortars (types 3and 4) versus that of the Portland cement mortars (type 5) with the gypsum stones and theanhydrite mortars (type 2).

Type of mortar Description

Type 1 Originalpietra rasa lime mortar (XIIIth C.), white

Type 2 Original anhydrite mortars (XIIIth C.), pinkish

Type 3Restoration Roman cement mortars (1896), thin pointing mortar, concave(spatuled), covering the original lime mortar, beige to greyish

Type 4Restoration Roman cement mortars (1898), thick and extruded repointingmortars, beige to greyish

Type 5Restoration Portland cement mortars (end of the XIXth C., as Type 3 and 4), castmortar to repair elements made of gypsum stone, dark grey

Type 6Restoration hybrid mortar (hydraulic lime and white cement) (1997-2003),

repointing deep joint, whiteTable 1 Types of original and restoration mortars identified on the faades

The use of anhydrite (CaSO4) or gypsum (CaSO4.2H2O), commonly with lime, to producejoint, precast or render mortars is reported in several historical buildings from the AncientEgypt period (e.g. Cheops and Unas pyramids [3]) to more recent periods of decorativearchitecture [4,5], through the Medieval period in Europe (e.g. North German [6] and Frenchcathedrals [7]). The choice of anhydrite or gypsum was usually motivated by the appearance(colour, texture) of hardened calcium sulphate based mortars close to that of lime mortar.Mixtures of lime and gypsum (so-called estrich gyps) are also reported in Germanmonuments.

Roman cement is a hydraulic binder developed in Europe in the early XIXth C. and widelyused both for civil engineering and architectural restoration applications. Several studies inthe field of stone and mortars conservation have revisited and highlighted the unique andlasting properties of this material to repair historical masonries [8-10]. Roman cements aresometimes called Natural cement, in contrast to artificial Portland cements (i.e. co-groundwith added gypsum) in some countries such as France [11] or USA [12]. This family ofhydraulic binder results from the calcination of naturally occurring limestones rich in clayminerals below the sintering point (800-1000C) and the grinding of the burnt material to afine cement. A typical shaft kiln was used during the early production of these cements.Roman cement has been produced from the 1830s in the North of Switzerland (Solothurn and

Aarau) and increasingly used throughout the XIXth C. The development of the Swiss railwaynetwork in the 1850s required rapid materials for the lines construction and contributed to thepromotion of the Swiss Roman cements. The annual production of cement reached 60 t in


14/41

1851. However, despite of the quality of the local raw materials, the Swiss production couldnot compete with the other European producers and the importation of Roman cement fromFrance increased from 1857 (with an import of 35000 tons recorded in 1876). The two main

production sites were Vicat (Grenoble) and Pavin de Lafarge (Virieu le Grand, Isre) [13].

3.Samples and methods

Figure 1 View of the Northern faade of the Valre castle and samples location (after Amsler andGagliardi)

Figure 1 gives the location of the mortars samples from the Northern faade of the church.The mortar 2 was sampled from the Eastern faade (annex buildings in ruins not illustrated in

Figure 1). Figure 2 gives detailed views (cross sections) of the four samples (three Romancement mortars and one Portland cement mortar) used for the microstructural characterisation(SEM and XRD). Scanning Electron Microscopy (SEM, Philips Quanta 200) was used tostudy the microstructure of the mortar samples. The samples were impregnated with epoxyresin and polished to obtain cross sections. The microanalysis of phases and elementalmappings were done with Energy Dispersive Spectroscopy (EDS, Bruker AXS Quantax).XRD analysis was done on sieved mortar samples using an XPert Pro PANAlyticaldiffractometer (Cu tube, =1.54 ). The Rietveld method was applied for the crystalline

phase quantification.

4.Results1. Macroscopic observations of the mortars samples

The Portland cement mortar (sample 1) was applied to replicate a column of a window frameoriginally made of white gypsum stone. From Figure 2, the bulk mortar sample looks beige(most probably due to the superficial atmospheric carbonation), but the cross section preparedfor microscopy reveals a grey matrix. The bulk mortar is dense and no specific degradation

pattern is observed. The outer subsurface (exposed to the environment) is notablydistinguished by a colour slightly switching from grey to light beige. The dissolution of thegypsum stone substrate, appearing yellow after epoxy resin impregnation, is well advanced

and a high amount of matter was lost during the sample preparation. A thick reaction interfaceis marked at the mortar/stone interface (top right image of Figure 2).


15/41

Figure 2 Photographs and description of the mortars samples

The bulk Roman cement samples have different colours according to their location and

application. Sample 2 was applied as a thin render on the original XIIIth C. anhydrite mortar.While the matrix of the cement mortar is dark grey (outer surface and bulk), that of theanhydrite mortar is pinkish and contains coarse white inclusions. The two mortars are strongly

bound by a very thin interface.Sample 3 is a thick joint partly covering and profiling an adjacent a gypsum stone (somefragments of stone remains visible in Figure 2). This sample was removed from the faadeand collected on the first roof of the Northern faade. The outer surface of the sample looks

beige but the cross section shows a dark grey matrix, comparable to the Portland cementsample 1. The cross section also reveals the formation of a thin (500 micron) and yellowishsubsurface that is exposed to the atmosphere. This specific layer was already reported inFrench [9] and Austrian [8] samples but its origin (atmospheric oxidation, footprint of organic

product, such as wax, originally used for technical or aesthetical purposes,) is not fully


16/41

understood. The mortar is composed of coarse aggregates. Micropores from mixing arerandomly distributed through the matrix.Sample 4 was collected from a core in a thick joint. The light colour (beige) of the bulkmortar remains even after the cross section preparation and this mortar is notably lighter thanthe previous ones. Figure 2 shows a dense matrix despite a high macroporosity formed during

the mixing.

2. XRD analysis on the mortarsThe crystalline composition of the RC mortars, suggested by the best-fit values of theRietveld analysis, is given in Table 2. The nomenclature of cement chemistry (C=CaO, S=SiO2, A= Al2O3, F= Fe2O3, C=CO3, H=H2O) is used for given phases in this table.

Origin of phase PhaseSample

1Sample

2Sample

3Sample

4Microcline KAlSi3O8 1.2 4.2 2.0 4.2

Albite NaAlSi3O8 14.3 15.0 5.7 16.4Clinochlore (Mg,Fe)[(OH)8AlSi3O8 3.7 5.6 3.5 4.7Muscovite K2Al4[(OH,Fe)AlSi3O10 4.0 3.6 5.0 3.0

(1)

Actinolite Ca2(Mg,Fe)5Si8O2.2(OH)2 2.2 2.5 1.6 1.6

(2) Quartz SiO2 17.7 13.9 16.8 13.0

Periclase MgO - - 1.4 -Tileyite Ca5(Si2O7)(CO3)2 - - 1.4 -Belite-2CaO.SiO2or -C2S 10.7 2.0 26.3 -Alite 3CaO.SiO2or C3S 7.4Gehlenite 2CaO.Al2O3.SiO2or C2AS 3.0 9.5 2.9 12.4

(3)

Ferrite Ca2(Al,Fe)2O5or C4AF 0.7 - 2.6 -(4) Calcite CaCO3or CC 9.0 26.2 15.3 35.4

Vaterite CaCO3 - 1.8 0.8 1.0(5) Aragonite CaCO3 - 6.2 - 6.2

(6) Portlandite Ca(OH)2or CH 7.5 - 1.7 -

Gypsum CaSO4.(H2O)2 3.6 8.4 3.4 2.0(7)Ettringite Ca6(Al(OH)6)2(SO4)3(H2O)26 4.4 - 8.4 -

Table 2 Quantitative XRD of the mortar samples : (1) from aggregates, (2) from aggregates or anhydrouscement, (3) from anhydrous cement, (4) from aggregates, anhydrous cement or carbonation product, (5)

carbonation products, (6) hydration products, (7) hydration products (Ett.) or reaction products withexternal sulfate

Table 2 discriminates different sources of crystalline phases. Some common minerals such asmicrocline, albite, clinochlore, muscovite and actinolite originate from the local sand and

were identified in various amounts in all samples. Quartz and calcite could also be attributedto the aggregate fraction in the Portland cement mortar (sample 1) but part of these phases can

be also attributed to residual remnants of Roman cements [14].

Despite the long time of contact with moisture on the faades, the roman cement samples (2 to4) contain unreacted phases. The polymorph of belite is identified in different amounts. Inthe typical range of calcination temperature of roman cement, -2CaO.SiO2dominates andhydrates afters few weeks of contact with mixing water and moisture [15]. When thecalcination temperature becomes higher, -2CaO.SiO2 can form but is less reactive than the polymorph. By achieving higher temperature in the kiln, more alumina from the raw clayminerals becomes available to form gehlenite (2CaO.Al

2O

3.SiO

2), a low reactivity phase.

Note that -belite and gehlenite are also identified in the Portland cement sample. Indeed -


17/41

belite is a secondary reactant in Portland cements much less reactive than tri-calcium silicate3CaO.SiO2, the main reactive phase responsible for the strength development.Gehlenite, which is sparingly present in contemporary Portland cements and has beenidentified in early PC (XIXth C.) [16], is said to be suggestive of underburnt Portland cement(usually calcined above 1400C) [17].

As abovementioned, the presence of calcite can be attributed to the mortar sand but also to thecarbonation of cement hydration products. Both Portland and Roman cement, and tri- anddicalcium silicate, respectively, hydrate to form calcium silicate hydrate (C-S-H) and calciumhydroxide (CH). Remaining calcium hydroxide was identified in samples 1 and 3 (Table 2)

but its carbonation under atmospheric conditions (e.g., CO2, wetting/drying cycles) led to theprecipitation of calcite and other metastable calcium carbonate polymorphs (aragonite andvaterite), particularly in the Roman cement samples. The metastable CaCO3 polymorphstransform to stable calcite by dissolution-precipitation reaction but can coexist in dryenvironment [18], which is the case of the Northern faade of the present structure. Note thatthe C-S-H phase could also be subject to carbonation and comparable carbonation products

could be attributed to this reaction.

Gypsum and ettringite were identified in the different mortars samples. Ettringite is a primaryhydration product in Portland cement (reaction between the calcium aluminate and the sulfateform added gypsum) but rapidly dissolve after the sulfate depletion (first days of the cementhydration). This phase can precipitate again when external or internal sources of sulfate areavailable. In the present case, this mortar was applied on a column originally made of gypsumstone, continuously providing sulfate allowing ettringite to precipitate. Some Roman cements(American Rosendale RC, French Vicat and Vassy RC) are reported to contain sulfate phases[9,19] leading to the early formation of ettringite. However ettringite was identified only inthe sample 3 (reprofiling mortar applied on the structural gypsum stone) while gypsum seemsto be the common reaction products present within this series of samples.

3. Microstructure of the mortars by SEMThe microstructure of the mortar samples is illustrated in Figure 3 to Figure 5.The main phases (typical to Portland cement) comprising a cement grain from sample 1(Figure 3.a) were identified as C3S (with Ca/Si= 3.15 0.3), C3A (with Al substituted by Fe,Ca/(Al+Fe) = 1.62 0.02) and C4AF (with Ca/Al = 2.09 0.05 and Ca/(Al+Fe) = 1.16 0.02). Note that C3S was not identified by the XRD technique. The observation of this sampleat lower magnification showed a many unreacted cement grains, suggesting a low degree of

hydration. The nature of the hydration products (around the cement grains) is discussedbelow.In sample 2 (Figure 3.b), three main phases were identified: C2S (rich in Si with Ca/Si= 1.77 0.2), C2AS rich in Si, Ca/Si = 1.7 0.16 and Ca/(Al+Si) = 0.76 0.05) and the silica gelappearing in dark in the BSE images, at the boundary of the cement grain. While C2S andC2AS are common phases formed in Roman cement at relatively high temperature ofcalcination [14], the presence of the silica gel at the boundary can be considered as aweathering product of the cement exposed to atmospheric carbonation. This product resultsfrom the reaction between C2S and CO2, leading to the decalcification of C2S and

precipitation of silica gel, which would be a residual product of Ca depletion. This specificreaction is scarcely reported but the mechanism and the nature of the silicate polymer were

recently studied on synthesized C2S and Portland cement [20].


18/41

Figure 3-c shows another example of Roman cement grain in which only rounded C 2S grainswere detected. Figure 3-d illustrates a Roman cement grain composed of C2S surrounded bysignificant amounts of silica gel (as carbonation product of parent C2S).

C3S

C3A

C4AF

C2S

C2AS

Silica gel

MgO

a: Sample 1 b: Sample 2

C2S

C2S

Silica gel

c: Sample 3 d: Sample 4Figure 3 Main features of the cement grains

Figure 4 illustrates the microstructure of the samples 1 and 2 and gives the elementalcomposition (Ca, Al, Si, S, expressed in the atomic ratio graphs) of the hydration products. Inthe Portland cement mortar (sample 1), we distinguish typically two types of calcium silicahydrate, the inner C-S-H immediately surrounding the cement grains, and the outer C-S-H

precipitated further in the microstructure. The latter are less dense and can incorporate other

phases such as ettringite or monosulfoaluminate. The EDS dots distribution in the atomic ratiograph suggests that outer C-S-H is intermixed with both ettringite (Ett.) and/ormonosulfoaluminate (Ms), while the latter was not detected by XRD, possibly because oftheir inherent poor crystallinity. The most important feature here lies in the comparison withthe composition of the hydration products of the roman cement mortar (sample 2). First, nodistinction between outer and inner products was observed in the hydrated binders. Inaddition, the atomic ratio graphs show a different phase assemblage than previouslydescribed. The distribution of Al, Ca and Si suggests that a C-A-S-H type phase intermixedwith calcium carbonate dominates in the microstructure, with no clear evidence of a pure end-member C-S-H phase. The degree of carbonation of this sample is very high (e.g., themicrograph of Figure 4 show a cement grain with fully carbonated C

2S) which probably

makes the analysis of outer hydration products difficult. However, recent studies on modernRoman cements [21] support the fact that C-S-H and CH, as hydration products of C2S, are


19/41

not homogeneously distributed in the microstructure even after 90 days of hydration, incontrast to Portland cements.The S/Ca vs.Al/Ca graph of the sample 2 shows that no sulphur bearing phase was detected,in the hydration products of the mortar.

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

0 0.2 0.4 0.6 0.8 1

Si/Ca

Al/Ca

innerCSH

outerCSH

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

0 0.1 0.2 0.3 0.4 0.5

Al/Ca

S/Ca

innerCSH

outerCSH

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

0 0.2 0.4 0.6 0.8 1

Si/Ca

Al/Ca

outerproducts

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

0 0.1 0.2 0.3 0.4 0.5

Al/Ca

S/Ca

outerproducts

Ms

Ett

C1.7SH

Ms

Ett

C1.7SH

CHorCC CHorCC

CHorCC

Ms

Ett

Ms

Ett

CHorCC

inner CSH

cement grain

outer CSH

+Ett.

cement grain

outer products

+CC

carbonated C2S

Sample 1

Al/Ca vs. Si/Ca

Sample 2

Al/Ca vs. Si/Ca

Sample 1

S/Ca vs.Al/Ca

Sample 2

S/Ca vs.Al/Ca

Figure 4 Microstructure and composition of hydration products, samples 1 and 2

The distribution of sulphur at the interface between the XIIth C. anhydrite mortar and theXIXth Roman cement mortar (sample 2) was studied by elemental mapping. Figure 5 shows aclear interface with no concentration gradient of sulphur at the boundary of the Romancement mortar. Nonetheless sulphur may diffuse and few dot are visible in the Figure 5-b,which were attributed to the reaction between sulphur and CH to precipitate gypsum in the

porosity.

Romancement

mortar

Anhydrite

mortar

a: BSE image b: Elemental distribution of sulfur

Figure 5 Elemental mapping of the interface between anhydrite mortar and RC mortar (sample 2)


20/41

Figure 6 shows BSE images and the respective segmented grey level images that illustrate thedistribution of the porosity (appearing in black) in the samples 2 and 4. In Figure 6- b and d,the pores resulting from mortar mixing are distinguished from the capillary porositydeveloped during the hydration of Roman cement. The network of the capillary pores israndomly distributed through the sample, allowing soluble sulphur to migrate and calcium

suphate (or ettringite) to precipitate in available space. This provides to Roman cementmortars some specific transfer properties adapted to the substrates (here, anhydrite mortar orgypsum stone).

a: BSE Sample 2 b: Segmented BSE (porosity) Sample 2

c: BSE Sample 4 d: Segmented BSE (porosity) Sample 4Figure 6 BSE images of samples 2 and 4 (a and c) and segmentation of grey level: distribution of the

porosity in the samples 2 and 4 (b and d)

5.ConclusionsThe use of Roman cement mortars in the late XIXthC. on the faades of the church of Valere(Sion, CH) was presented. The specific application as joints of structural gypsum stones andrender of earlier mortar made of anhydrite was discussed. Roman cement belongs to a specificfamily of hydraulic binders developed during the XVIIIth C and which shows interestingcompatibility properties adapted to stone conservation.The samples show a grade of cement containing significant amounts of unreacted gehleniteand -C2S, suggesting a calcination process using a temperature range above 1000C, likelythat of the French Vicat cement and higher than typically reported for Roman cements (800-

1000C). Recent studies showed that the microstructure development of hydrated modernRoman cements is controlled by dissolution of amorphous CaO-Al2O3 and fine calcium


21/41

carbonate (both from the raw cement) and precipitation of carbonated phases in the CaO-Al2O3-CO3-H2O system. If sulphur is present in the raw cement, ettringite andmonocarboalumiante precipitate, as in the case of the French Vicat cement containing gypsum[19,21]. These elongated crystals (plate of carbonated phases or ettringite needles) precipitatevery rapidly to form a poorly packed microstructure in the first minutes of hydration.

However, this phase assemblage develops minimal strength, allowing the rapid application ofmortars for cast elements or renders. This microstructure becomes progressively and locallydenser after the later reaction of calcium silicate (mainly -C2S but also -C2S for thecements produced at higher temperature) forming, in theory, C-S-H and CH. However the

presence and location of these two phases in hydrated Roman cement remains difficult toidentify in contrast to the case of C3S hydration in Portland cements. This identification

becomes more challenging in highly carbonated samples, where different calcium carbonatepolymorphs are uniformly distributed in the primary hydration products. Carbonation actsalso on the non reacted C2S, during which decalcification leads to the formation of silica gelat the cement grain boundary. This reaction may contribute to the reduction of the extent ofC2S dissolution and cement hydration.

The excellent state of conservation of Roman cement mortar could be explained by itsmicrostructure and the high capillary porosity developed during hydration [22,23]. Thisallows the sulphur, dissolved from the gypsum stone, to migrate through the Roman cementmortar and to potentially react to form gypsum (or ettringite in sample 3) without generatinginternal stress and subsequent cracks formation. This specific feature is one of the mostimportant characteristics that make Roman cements highly suitable as a restoration mortar incontrast to the Portland cement mortar that was inappropriate for replicating a windowcolumn originally made of gypsum stone.

6.AcknowledgementsThe European Project ROCARE is acknowledged for the financial support of this case study.

The architect C. Amlser is also gratefully acknowledged for his interest in this material andthe current opportunity to use the Church faade as a demonstration site for modern Romancement mortar applications. CG would like to dedicate this paper to the memory of Pr.Michele Coutant, her knowledge and her passion of the conservation of cultural heritage.

7.References[1] Raemy-Berthod, C., 2003. Inventaire suisse darchitecture, 1850-1920: Sion. Socit

dHistoire de lArt en Suisse9 92.[2] Winkler, E.M., 1997. Stone in Architecture: Properties, Durability. Springer[3] Regourd, M., Kerisel, J., Deletie, P. and Haguenauer, B., 1988. Microstructure of

mortars from three Egyptian pyramids. Cement and Concrete Research18(1) 81-90.[4] Cotrim, H.l., Veiga, M.d.R.r. and de Brito, J., 2008. Freixo palace: Rehabilitation of

decorative gypsum plasters. Construction and Building Materials22(1) 41-49.[5] Luxan, M.P., Dorrego, F. and Laborde, A., 1995. Ancient gypsum mortars from St.

Engracia (Zaragoza, Spain): Characterization. Identification of additives andtreatments. Cement and Concrete Research25(8) 1755-1765.

[6] Livingston, R.A., A. Wolde-Tinsae and Chaturbahai, A., 1991. The Use of GypsumMortar in Historic Buildings. In: C. Brebbia, Structural Repair and Maintenance of

Historic Buildings, Southampton UK,

[7] Adams, J., Kneller, W. and Dollimore, D., 1992. Thermal analysis (TA) of lime- andgypsum-based medieval mortars. Thermochimica Acta211(0) 93-106.


22/41

[8] Weber, J., Gadermayr, N., Bayer, K., Hughes, D., Kozlowski, R., Stillhammerova, M.,Ullrich, D. and Vyskocilova, R., 2008. Roman cement mortars in Europe'sarchitectural heritage of the 19th century.ASTM Special Technical Publication, 69-83.

[9] Gosselin, C., Verges-Belmin, V., Royer, A. and Martinet, G., 2009. Natural cementand monumental restoration.Materials and Structures/Materiaux et Constructions42

(6) 749-763.[10] Hughes, D.C., Swann, S. and Gardner, A., 2007. Roman cement: Part one: Its originsand properties.Journal of Architectural Conservation13(1) 21-36.

[11] Avenier, C., Rosier, B. and Sommain, D., 2007. Ciment naturel. Glnat[12] Werner, D. and Burmeister, K., 2007. An Overview of the History and Economic

Geology of the Natural Cement Industry at Rosendale, Ulster County, New York.Journal of ASTM International4(6) 14.

[13] Dariz, P., 2009. Romanzement in der Schweiz Geschichte des natrlichhydraulischen Bindemittels in der Eidgenossenschaft.Restauro 8 8.

[14] Hughes, D.C., Jaglin, D., Kozlowski, R. and Mucha, D., 2009. Roman cements - Belitecements calcined at low temperature. Cement and Concrete Research39(2) 77-89.

[15] Hughes, D.C., Jaglin, D., Kozlowski, R., Mayr, N., Mucha, D. and Weber, J., 2007.Calcination of marls to produce Roman cement.Journal of ASTM International4(1)

[16] Pinter, F.,personnal communication: Vienna.[17] Campbell, D.H., 1999. Microscopical examination and interpretation of portland

cement and clinker. Portland Cement Association[18] Thiery, M., Villain, G., Dangla, P. and Platret, G., 2007. Investigation of the

carbonation front shape on cementitious materials: Effects of the chemical kinetics.Cement and Concrete Research37(7) 1047-1058.

[19] Vyskocilova, R., Schwarz, W., Mucha, D., Hughes, D., Kozlowski, R. and Weber, J.,2008. Hydration processes in pastes of Roman and American natural cements.ASTMSpecial Technical Publication, 96-104.

[20] Shtepenko, O., Hills, C., Brough, A. and Thomas, M., 2006. The effect of carbondioxide on -dicalcium silicate and Portland cement. Chemical Engineering Journal118(1-2) 107-118.

[21] Gosselin, C., Scrivener, K.L. and Feldman, S.B., 2011. Microstructure of romancements used for architectural restoration. International Conference on CementChemistry, Madrid,

[22] Klisinska-Kopacz, A., Tislova, R., Adamski, G. and Kozlowski, R., 2010. Porestructure of historic and repair Roman cement mortars to establish their compatibility.

Journal of Cultural Heritage11(4) 404-410.[23] Bayer, K., Gosselin, C., Hilbert, G. and Weber, J., 2011. Microstructure of historic and

modern Roman cements to understand their specific properties. In: T.K. AlenkaMauko, Tinkara Kopar, Nina Gartner, 13th Euroseminar On Microscopy Applied ToBuilding Materials, Ljubljana, 2-3.


23/41

IDENTIFICATION OF 19th

CENTURY ROMAN CEMENTS BY THE PHASE

COMPOSITION OF CLINKER RESIDUES IN HISTORIC MORTARS

Nina Gadermayr1, Farkas Pintr2, Johannes Weber1

1University of Applied Arts Vienna, Institute of Arts and Technology, Section of

Conservation Sciences

A-1013 Vienna, Salzgries 14/1, Austria2Federal Office for the Protection of Monuments, Scientific Laboratory,

A-1030 Vienna, Arsenal 15/4, Austria

AbstractThe term cement in its historic context stands for several groups of highly

hydraulic binders which may differ in the temperatures of calcination and hence yieldmortars of significantly differing properties. A key to identify the exact type of binderused in a historic cement mortar consists in the textural and mineralogical features ofunhydrated residual clinker particles. Employing various techniques of light andscanning electron microscopy allows define fingerprints for given cements, a task hardlyaddressed in earlier studies on natural cements calcined at low temperature.

The present contribution focuses on the most characteristic residual clinker phasesfound in historic Roman cement mortars, a group of materials widely used in the 19thcentury urban architecture and civil engineering. These binders were produced fromnatural marlstones through shaft kiln calcination at temperatures virtually below

sintering. High amounts of non-crystalline reactive compounds form in this lowtemperature regime. Along with fine-grained impure C2S, they have developed into arelatively homogeneous hydrated matrix. In parallel, however, non- or low reactivecompounds have assembled together in binder-related nodules characteristic for Romancement mortars. Most of these compounds are of non-crystalline nature, they comprisesolid solution silicate phases as well as crystalline CS (wollastonite), coarse C2S andC2AS (gehlenite). By means of thin section and reflected light microscopy, combinedwith scanning electron microscopy and X-ray microanalysis, the phase assemblages canbe observed and identified in virtually all of the historic mortars. Various classes ofresidues were distinguished within the above general frames. According to their textureand mineral content, they are classified into overfired or underfired in respect to the

optimum temperature of calcination. Optimum clinker assemblages with high amountsof reactive phases may form nodules as well, characterised by their specifically densehydrate structure.

Following a general description of the above mentioned classes, the contributionpresents a group of samples from historic faades in the city of Budapest as an exampleof the usefulness of the classification in the practice of mortar analysis. The data suggestthat different brands of Roman cements characterised by specific setting times may haveexisted and used on purpose for specific mortar applications.

The paper aims at providing information useful for the identification andcharacterisation of historic natural mortars.

Keywords:Roman cement, natural cement, phenograins, clinker microscopy, SEM


24/41

1. IntroductionNatural cements manufactured by the low temperature calcination of marlstones

fine-grained carbonate rocks with significant amounts of clay and other silicates wereamongst the most important building materials of the 19 thcentury in Europe. Almost a

century after their recession in the first half of the 20 thcentury and following a period ofneglect, these binders, frequently referred to as Roman cements (RCs), are nowattracting new interest by the restoration community. This development wasaccompanied by a number of scientific and technical studies on the composition andproperties of the cements, pastes and mortars. Amongst other activities, the EU-fundedprojects ROCEM (2003-06) and ROCARE (2009-12) have provided detailedinformation which can be viewed e.g. in the project website www.rocare.eu.

A number of handbooks and other texts edited in the 19thand early 20th centuriesdeal with the technical aspects of the Roman cement technology, e.g. Pasley (1830),Tarnawski (1887), Tetmajer (1893), Schoch (1904), Eckel (1905), Bohnagen (1914),

Khl & Knothe (1915). Analyses of the chemical composition of RCs, descriptions ofthe process of production, and definitions of their key properties can be found in thesesources. Historic standards took account of the typically short setting time of Romancements, which were less than 15 minutes for many brands. Evolvement of the finalstrength is generally delayed in comparison to Portland cements (PCs), so that the 28-days strength of the former is of not too much significance for its quality. Concerningthe temperature of calcination, this was established by trial calcinations yielding binderswhich match the historic ones by all means (Hughes et al., 2007, 2009). Thus, thetemperature needed to produce optimum Roman cements from an appropriate raw feedaverages at or even below about 900 C. At this temperature, the mineralogicalcomposition of a RC analysed by X-ray diffraction is characterised by a specific phaseassemblage.

According to Hughes et al. (2010, quoted verbatim in the following), optimalcements are characterised by maximum -belite content, a high content of anamorphous phase and residual calcite and quartz indicating incomplete calcination;Carbonated belite, spurrite, was observed in some cements at low temperatures. As thecalcination temperature is increased, the -belite converts to the -belite form observedin natural hydraulic limes and Portland cement; spurrite was also reduced. Additionally,aluminosilicate as gehlenite is observed. These developments are accompanied by areduction in the amorphous phase and residual calcite and quartz. Brownmillerite wasobserved in cements with a high iron content. (End of quotation).

When dealing with the identification of historic RCs, it must be kept in mind that,being natural cements from different sources, they show a relatively wide spread of theirchemical composition, sometimes falling within the much narrower range of PCs. Thisfact, as well as the well-known problems related in general to the analysis of binders inhistorical mortar samples, makes it difficult to identify a Roman cement mortar by justthe chemical composition of the binder. Also X-ray diffraction can be of just a limiteduse, since fingerprint products like -belite have reacted away by hydration to C-S-H; similar holds for the reactive portion of the amorphous phase, which is anyway notdefinable by this method. It is thus due to a petrographic analysis based on methods ofmicroscopy to assess the characteristic features of a historic Roman cement mortar, at

least in cases where the experienced eye of an expert fails to take an unambiguousdecision on the nature of the mortar.


25/41

Studying the compounds in the way outlined above means to observe and analysethe residual cement clinker in an otherwise hydrated and frequently carbonated mortarsample, by all their petrographic and mineralogical properties accessible to the light andelectron microscope. This approach forms the methodical base of the present study

which thus aims to contribute to a better knowledge of characteristic fingerprints for theidentification of Roman cement in historic mortars. Beyond this, it will be shown that amore precise classification of the type of RC can be achieved.

2. MethodsFollowing vacuum-impregnation in epoxy resin, the mortar samples were cutperpendicular to the surface and processed to petrographic thin sections. They were thenstudied under a polarising microscope (PL) in transmitted light, occasionally also inreflective light in the case of polished thin sections. The scanning electron (SEM)studies were performed on polished thin sections, or on their counterpart sections,

respectively, employing a back-scattered electron detector (BSE). Usually, the high-vacuum working mode was selected with prior carbon coating of the surfaces, though anumber of analyses were also made on uncoated samples in the low-vacuum mode.Chemical analyses were performed by an energy-dispersive X-ray analyser (EDS)linked to the SEM.

3. Phenomenology of residual clinker in historic Roman cement mortarsThe most characteristic common features of all historic RC mortars are the binder-

related nodules, phenograins according to the nomenclature by Diamond & Bonen(1993), which are formed by various types of incompletely hydrated cement particles ordensely hydrated agglomerates, respectively. They can be observed in any microscopeeven at low resolution, though determination of their specific nature needs more precisemethods of analysis such as polarising microscopy on thin sections combined withelectron microscopy.

The abundant presence of such nodules is believed to be due to the particularities ofthe historic process of production, see e.g. Weber, Gadermayr, Bayer et al. (2007): Theraw feed, a rock material with all its natural impurities and inhomogeneities, enters thevertical shaft kiln unground in fist-sized fragments. The temperature gradients in such akiln are paralleled by gradients within every single lump of stone, so that the resultingclinker is likely to cover a wide range of different grades of calcination. The lowtemperatures below effective sintering even favour uneven conditions of reaction.

Notwithstanding the occasionally reported fact that obviously over- or underburnedmaterial was removed manually before the clinker was ground (Khl and Knothe 1915),the final product is a blend of differently reacted cement grains which, as a consequence,vary in respect to their reactivity with water to form hydrates.

Given the generally coarse particle size of a historic RC frequently in the range ofseveral hundreds of micrometers up to one millimetre, those clinker particles with nonor incomplete hydration remain well visible to the eye of the observer. They bearsignificant information on the historic process of production as well as on the nature ofthe raw feed. In earlier studies performed on Roman cements calcined at definedtemperatures in the laboratory, a classification of these phenograins was attempted inrespect to the strength development of the corresponding pastes (Weber, Gadermayr,Kozowski et al., 2007). Reflecting their respective grades of calcination, the residual


26/41

clinker can be classified roughly as either underfired(1), optimally fired(2) oroverfired(3), even if the boundaries between the groups are fluid to some extent.Residual cement nodules can be observed even in carbonated Roman cement mortars, -as noted by St.John, Poole and Sims (1998) - and classified by their structure.

3.1 Type 1 nodules underfiredUnderfired nodules calcined to a sub-optimal degree, herewith referred to as type 1

nodules, still show some textural features inherited from the marl (Figures 1a+b):Unreacted silicates such as predominantly quartz, mica and feldspar can be observedembedded in a flaky to fibrous matrix of Ca-Al-silicates. Partial diffusion of Ca and/orK into silicate minerals can be traced by very thin marginal zoning visible by SEM-BSE,which is in general better developed in quartz grains than in feldspars, but can be alsoobserved in coarse micas. It is especially the quartz which serves as indicator forunderfired conditions, as it appears angular and with a still intact crystal lattice, showing

the typical birefringence visible under crossed polars in PL, with no or only very thindiffusion rims of Ca or K to be observed by SEM.Calcite and the matrix clay minerals appear to be the first components to react at

elevated temperatures. Calcite is incompletely decomposed, however, and often formspseudomorphs after fossil shells. These features related to calcite are another significantcriterion for sub-optimal calcination. The texture of the clayey matrix, especially inrespect to the orientation and porosity of the constituents, forms a third sensitiveindicator of the conditions of calcination. Compared to optimally fired state, thematrix of underfired nodules principally differs by its relatively dense structure near thelimit of resolution of the SEM.

Figure 1a, b. Typical underfired residual grain (type 1). (a): SEM-BSE image, (b): the same particlein the polarising microscope under crossed polars; a characteristic quartz grain can be seen right fromthe centre of the micrograph.

3.2 Type 2 nodules optimally firedThe common feature of these optimally calcined residues is their relatively high

amount of reactive clinker material. Thus, apart from the composition and texture of theunreacted portion, a significant amount of hydration products is visible either as acompact rim of varying thickness up to 150 m around the nodules, or even in the formof nearly completely hydrated nodules which, however, include unreacted phase

residues. Such densely hydrated areas may have uncarbonated core areas even inotherwise fully carbonated mortars. According to their characteristic constituents, the

a b


27/41

optimally fired nodules can be subdivided into two types; they contain either non-crystalline silicates with clear zoning (2-A), or- belite clusters (2-B), respectively.

Residues of the type 2-A (Figures 2a+b, 3a+b, 4a+b) are characterised by non-equilibrium features such as solid solution systems, and by significant zoning due to a

partial diffusion of Ca predominantly from carbonates, and K probably from micas -into silicates. The most common phases are non-crystalline silica with no birefringence,up to a size of 30 m. Varying amounts of K and Ca are always present, by theirdiffusion into the silicate they have supposedly contributed to the breakdown of theircrystalline lattice. Since K-ions have diffused more easily, this element appears to beconcentrated rather in core areas of zoned silicate grains, where eventually it may havecaused partial melting visibly by small spherical cavities. Ca, on the other hand, tends tobe concentrated rather in the marginal zones, where it may form calcium silicates ofstoichiometric composition. Thus, the front of Ca-diffusion into silicates can beobserved by a sequence of Ca-silicate minerals with increasing Ca:Si ratios, such as

wollastonite CS, rankinite C3S2and belite C2S.Where no full hydration of the clinker matrix has occurred, it consists of very fineCa-Al-Si-phases at the limits of resolution. These non-crystalline products of calcinationof the clayey portion of the raw feed play a dominant role in the early age hydration of aRC (Tilov 2008). Depending on the specific grade of calcination, the residual matrixappears either as densely fibrous, or as porous and grainy. New formations within thematrix include gehlenite C2AS along with the Ca-silicate minerals mentioned above.

Figure 2a, b. Characteristicdensely hydrated nodule of the 2-A type. (a): SEM-BSE image, note theshrinkage cracks, (b): the same particle in the polarising microscope under crossed polars.

Figure 3a, b.Type 2-A typical optimally fired phenograin; (a) SEM-BSE of the whole nodule,

(b) detail of (a) with wollastonite CS and belite C2S grown around silicate grains.

a b

a b


28/41

Figure 4a, b.SEM-BSE images of a optimally fired clinker nodule, (b): detail of the same clinkerresidue: crystallisation of C2S at the costs of zoned silicate grains this grain could be classified astransition between types 2-A and 2-B: Belite C2Sis growing around silicates and in form of smallerparticles in a more flaky matrix.

Type 2-B-residues (Figure 5a+b) are characterised by distinctive clusters ofrelatively coarse belite C2S, associated with minor amounts of phases like C3S2, CS,C2AS, and occasionally Ca-Fe-Ti-phases.

These belite crystals of a size up to 8 m show lamellar structures and equilibriumfeatures such as triple borders. They may contain small amounts of Mg and wereobviously formed by a full solid state reaction with the silicates. The crystals arefrequently arranged around open spaces probably formed by local melting. As comparedto similar voids found in type 2-A clinker nodules, the voids of 2-B tend to be bigger, i.e.up to 80 m in diameter, and their margins are defined by euhedral clinker crystals.

The coarse nature of belites prevents them from efficient hydration. Thus, usuallyjust a dense rim of hydrates can be found around such clusters.

Figure 5a, b.SEM-BSE micrographs of a remnant of type 2-B: clusters of belite. (a): note the denserim of hydration; (b): detail of (a), with euhedral belites (light grey) and rankinite, C3S2(dark grey).

3.3 Type 3 nodules overfiredOverfired super-optimal particles (Figures 6a-d) show no or just weak signs of

hydraulic activity. They are easily recognised by their very angular and often shard-likestructure. Marginal hydration of type 3 nodules can never be observed, though in thecase of small particle size their hydration can be assumed.

a b

a

b


29/41

Due to different cooling rates in the kiln during the manufacturing process, twoforms of overfired particles can be observed in RC: Angular shaped and coarsecrystalline (type 3-A) or shard-like and glassy relicts (type 3-B) with air voids.

Abundant compounds of the coarse crystalline nodules (type 3-A, Figure 6a+b) are

gehlenite C2AS, belite C2S, wollastonite CS, leucite S2AK, and various phases of theCaO-SiO2-Al2O3-K2O-FeO-TiO2 system. The resulting compounds are not only well-defined stoichiometrically, but also in terms of their euhedral crystal habits. Thus, it canbe assumed that they were formed under equilibrium conditions.

Glassy nodules (type 3-B, Figures 6c+d) show various stoichiometries in the Si-Al-K-Ca system, though in general they contain significant amounts of aluminium. In theisotropic glass, the presence of small microlithes indicates devitrification.

Overfired nodules probably formed in local areas of higher temperatures in thetraditional kilns seem to play an important role as inert filler, also controlling the rapidsetting of RC to some extent (Hughes et al., 2010). Generally, this mixture of various

residues is consistent with results of laboratory calcinations of marls, where optimal,sub- and super-optimal calcined cements show similar microstructures (Hughes et al.,2010).

Figure 6a-d.SEM-BSE micrographs of overfired phenograins. (a, b): Type 3-A is characterisedby its coarse-crystalline structure. (b): detail of the same nodule: C2S (medium grey), wollastoniteCS, a mineral of the composition C2S2(A,M) - probably melilite (medium to dark grey), kalsiliteKAS (dark), Ca-Ti-Fe-mineral (bright).(c, d): Type 3-B: Si-Al-K-rich glass with small microlithes and bubbles; details of the nodule in (d) -

the darker parts are richer in Si.4. Roman cement clinker residuals in historic mortars from Budapest

a b

c d


30/41

Samples collected from seven 19thcentury apartment houses in Budapest (Figure 6)were selected to check the above described typology of RC clinkers in the course of fullmortar analyses. It was hoped that some kind of correlation between the predominanttype of clinker nodules and the way of architectural application could be established. A

list of samples is given in Table 1.Approximately 100 phenograins were analysed and statistically evaluated for each

of the samples. They were classified into one of the three major phenograin types 1, 2 or3 described in Section 3 of this study.

In all of the samples, the optimally fired cement grains revealed to dominate (55 to66%). The amount of under- and overfired nodules varies from 8 to 28% and from 15 to32%, respectively (Table 1). The samples can be divided into two groups based on theratio of under- to overfired particles in a mortar (Figure 7): one group, comprising allrender samples, contains higher amounts of overfired nodules; the other one,representing all pre-cast elements, shows higher amounts of underfired phenograins.

Based on the statistical evaluation of the observations, the following assumptionscan be made. .The high amount of optimally fired cement grains in all samples is likelyto indicate a pre-selection of optimally calcined clinker material before the grindingprocedure. Furthermore, as suggested by Weber, Gadermayr, Kozowski et al. (2007),the overfired portion of a Roman cement clinker would hydrate more slowly and thuscontribute to a prolonged setting time of a mortar. Given the generally short times ofworkability of Roman cements, retardation is a prerequisite for on-site render works.Therefore, the higher amount of overfired cement grains in the three in-situ run samplessuggest slower setting times of theseRoman cement mortars.

In contrast to the above, in the other four samples the predominance of optimallyand underfired clinkers suggests a faster speed of set of the mortars (Weber, Gadermayr,Kozowski et al., 2007). Since these mortars were also used for render works whereprolonged setting times were required, another method of retardation must have beenused. A simple method recorded by historic sources was e.g. the storage of Romancement clinker in the open air for some days. Under these conditions, a certain portionof highly reactive and amorphous Ca-aluminate phases of the clinker would react withthe moisture of the air, thus withdrawn from the early hydration responsible for themortar set. Consequently, a slower speed of set and better workability could have beenachieved.

Table 1. Mortar samples taken from 19thcentury building faades in Budapest with quantitative

classification of the phenograins.

Sample Type of sampled faade element

ROC-1 In-situ runRC mortar, door framing

ROC-2 RC mortarfrom diamond-shaped element

ROC-3 RC mortarfrom diamond-shaped projection

ROC-4 In-situ runRC mortar, window framing

ROC-5 Mortar, quoin element,

ROC-6 In-situ runRC mortar profile

ROC-7 Mortar, RC-based artificial stone column


31/41

Figure 6. Characteristic apartment housefrom 1896 in downtown Budapest.

Figure 7. Quantitative evaluation of phenograins inhistoric RC mortar samples from Budapest

5. Conclusions and discussionThe selected approach to study the clinker residues in historic mortars arose from

the fact that neither chemical nor X-ray diffraction analyses could yield sufficientinformation to identify natural cements of the Roman cement type versus e.g. Portlandcements or highly hydraulic limes. Another objective was related to the search foroptimum conditions of calcination in order to reproduce Roman cements matching thehistoric binders in their properties and composition. To that end, the residual clinkers in

historic mortars needed to be studied in comparison with the new products.In general, the variety of phases and textures present in Roman cement phenograinswithin even one sample is striking. The basic phenomenology and classification of themost frequent types of nodules found in Roman cement mortars, as presented in thiscontribution, is based on observations made for a considerable number of samples frommany European sources. One should therefore keep in mind that the describedphenomena are just indicative in a way that significant deviations can occur for certainRoman cements from specific regions. To give just two examples, clinkers found in RCbrands produced from dolomitic marl, e.g., could not be considered in this contribution,for the simple fact that so far the authors have not come across them frequently enough.Some historic RC brands from England, manufactured from septaria by using coal as a

fuel rather than timber, show distinct features related to higher temperatures and sometimes higher sulphate contents inherited from the stone. The related clinkerassemblages were not included in this study.

The classification into different types of phenograins reflecting various grades ofcalcination forms a promising approach to gain more insight into mechanisms ofhydration on the one hand, on the other hand it will enable improved differentiationbetween RC brands once a more systematic assessment will be made.

It is hoped that increasing numbers of scientist will use the approach presented topublish comparable observations and data on RC mortars from all over Europe, so thatour knowledge on the range of composition within this family of binders could grow.


32/41

Acknowledgements

The present study is largely based on research activities funded by the Commissionof the European Union within the EU research projects ROCEM (2003-06) and

ROCARE (2009-12).

6. ReferencesBohnagen, A. 1914. Der Stukkateur und Gipser. Reprint Verlag LeipzigDiamond, S. and Bonen, D. 1993. Microstructure of Hardened Cement Paste A newInterpretation,Journal of the America Ceramic Society, 76(12): 2993-99.Eckel, E. C. (1905): Cements, Limes and Plasters 1st ed. New York: John Wiley &Sons Inc.

Hughes, D., Jaglin, D., Kozlowski, R. et al. 2007. Calcination of marl to produceRoman cement. Journal of ASTM International, 4(1), Paper ID JAI100661.Hughes, D.C., Jaglin, D., Kozowski R., et al. 2009. Roman cements - Belite cementscalcined at low temperature. Cement and Concrete Research 39, 7789.Hughes, D., Weber, J., Kozlowski, R. 2010. Roman Cement for the Production ofConservation Mortars. Preprints 2nd Historic Mortars Conference & Rilem TC 203-RHM Repair Mortars for Historic Masonry Final Workshop, Prague, 22-24.September2010.Khl, H. and Knothe, W. 1915. Die Chemie der hydraulischen Bindemittel Wesenund Herstellung der hydraulischen Bindemittel. Leipzig: Verlag v. S. Hirzel.St. John, D. A., Poole, A. W., Sims, I. 1998. Concrete petrography: A handbook ofinvestigative techniques London etc.: Arnold.Pasley, C. W. 1830. Observations, deduced from experiment, upon the natural watercements of England, and on the artificial cements, that may be used as substitutes forthem. Printed by authority, at the Establishment for Field Instruction.Schoch, C. 1904. Die moderne Aufbereitung der Mrtel-Materialien. 2. Aufl., Berlin:Verlag der Thonindustrie-Zeitung.Tarnawski, A. 1887. Kalk, Gyps, Cementkalk und Portland-Cement in sterreich-Ungarn. Wien: Selbstverlag.Tetmajer, L. 1893. Methoden und Resultate der Prfung der Hydraulischen Bindemittel.Mitteilungen der Anstalt zur Prfung von Baumaterialien am eidgen. Polytechnikum

Zrich. 6. Heft, Zrich: Selbst-Verlag der eidgen. Festigkeits-Anstalt.Tilov, R. 2008. Hydration of natural cements. Ph.D. dissertation. Institute ofCatalysis and Surface Chemistry, Polish Academy of Sciences.Weber, J., Gadermayr, N., Bayer, K. et al. 2007. Roman cement mortars in Europesarchitectural heritage of the 19th century. Journal of ASTM International, 4(8), PaperID JAI100667.Weber, J., Gadermayr, N., Kozowski, R. et al. 2007. Microstructure and mineralcomposition of Roman cem

Documents

Oral Presentations—Methods and Materials of Cleaning, Conservation, Repair and Maintenance; Logistics and Planning